A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.
This invention relates to optical sensor systems for determining the location or orientation of an object, such as an electronic component, including such systems which report a physical condition of the object, such as the orientation of the object, the height of a particular feature on the object, the distance between features on the object, the presence of expected features on the object or the coplanarity of features on the object.
There are many optical-based approaches to identifying various parameters related to an object, such as an electronic component. The various parameters which may be reported include orientation of the component, location of features on the component, indications of the quality of the component, such as a bent lead (lead tweeze) in the case of a Quad Flat Pack (QFP) component, or a measurement of a feature located on the bottom of the component, such as the height of a ball on a Ball Grid Array (BGA) Component. The system for identifying these parameters is generally mounted in a pick and place machine which picks up the component and places the component on a printed circuit board. All of these approaches rely on a quill in the pick and place machine to pick up the component, and all generally utilize the motion control system in the pick and place machine in the measurement process. Some of the approaches use cameras to collect data representative of an image of the component and others collect data representative of an outline of the component, and all have some sort of processor that analyzes the collected data in order to identify some specific portion of the component, such as the edges of the part or the tips of the component leads. The sections below briefly describe a variety of the approaches used today, and describe their relative advantages and disadvantages.
In general, the systems described below can be characterized as being either xe2x80x9con-headxe2x80x9d or xe2x80x9coff-head.xe2x80x9d On-head sensor systems (considered together with their host pick and place machines) sense and correct for the orientation of the component while the component travels to the target circuit board. On-head systems are preferred, as they minimize the amount of time required to place a component on a board, thereby increasing the throughput of the pick and place machine.
Off-head sensor systems perform the data collection in a fixed location on the pick and place machine. With an off-head system, the quill must first move the component to the dedicated station and after acquiring the image, move the component to the target circuit board for placement. Off-head systems typically reduce the machine""s throughput since the head makes additional stops or travels further, but are used because they are inexpensive, reduce the mass which must be carried on the head and one sensor can readily service multiples quills.
Another way to characterize various sensor systems is by the relative location of the detector and the light source with respect to the component. FIGS. 1A-C show the three schemes of front-lit, back lit and shadow cast sensor systems, respectively. In front-lit systems, light from a source illuminates the features of interest on the component and the light reflected off of the features is optionally directed through optics forming an image on a detector. In back-lit systems, light from a source is incident on the component, and then optionally directed through focusing optics before it reaches a detector. Because the detector does not collect light scattered off the bottom of the component in either the back-lit or the shadow-cast systems, neither type of system is capable of inspecting features on the bottom of components, such as balls on a Ball Grid Array (BGA) or pins on a pin grid array (PGA). In shadow cast systems, light from a source optionally travels through collimating optics, onto the component, optionally directed through an optical system and then casts a shadow on the detector.
One approach in use today for assessing the orientation of an electronic component on a pick and place nozzle uses data representative of a shadow of the component, sometimes exclusively using the data representative of the edges of the shadow to perform its orientation computation. One such sensor is available from CyberOptics(copyright) Corporation and is called a LaserAlign(copyright) Sensor. In this shadow cast sensor system, light is cast onto the component from one side of the component and a linear detector on the other side of the component captures data representative of the component""s shadow. Appropriate electronics analyze the data to compute the x,y,xcex8 coordinates of a pre-selected feature on the component. FIG. 2 shows this approach schematically and the method is described further in U.S. Pat. No. 5,278,634 to Skunes et al., assigned to CyberOptics Corporation. This approach takes advantage of the motion control system already installed on the pick and place machine, which has an accurate encoder for reporting the angular position of the nozzle. At several angular positions, the system collects data from the detector representative of the shadow of the component. As appropriate for the type of component, the angular positions do not have to be equally spaced over 360 degrees nor is the component required to spin at a constant angular velocity. For each angular position, the sensor system computes the width of the component""s shadow. By assessing the shadow center position at the minimum shadow width angle the component location in one axis can be determined. By measuring two shadow positions at the two minimum widths 90xc2x0 apart, the position of a rectangular component can be computed. This system does not provide any further information not already derived from the shadow edges. Hence, it cannot identify misplaced or missing balls on BGAs or flip-chips. One main advantage of the sensor, however, is that it can be placed on-head, which allows for increased throughput of the pick and place machine.
Another on-head system employs a video camera on the moving head. In this approach, the camera stares down parallel to the pickup quill, and a right angle prism (or equivalent) is moved under the camera and component during travel, shown in FIGS. 3A-B. In this approach, there is typically one camera for each pick-up nozzle. This system has three drawbacks. First, there is substantial cost and mass associated with the right angle prism and the mechanism to slide it reliably into place. Second the minimum time between pickup and placement is limited by the time required to raise the nozzle, slide the prism into position, acquire the image, slide the prism out of position, and then lower the nozzle. Finally, the prism must be roughly as thick as the width of the component to be measured. Hence, the z motion of the component during pickup and placement is much longer than is required for other approaches.
One off-head approach to assessing component orientation is to place one or more fixed position upward looking video cameras (e.g., 2D array detectors) in the pick and place machine at a dedicated station. For small components, or ones with relatively coarse pitch between leads, one image of the component is taken and its position determined from analyzing a single image. For larger components with fine pitch, several images may be needed to provide adequate resolution, since the component is placed on the basis of the location of its feature positions and a higher magnification view may be required to resolve fine pitch leads, balls, columns or pins. This approach utilizes standard components (e.g., cameras, optics, image processing hardware and image processing software) which may be easily assembled by pick and place machine manufacturers. However, this off-head system is slower than on-head approaches for the reasons stated above. The path from pickup to placement of the component must be altered to travel over the camera, degrading overall throughput of the pick and place machine. It is also likely that the travel will need to be slowed or stopped over the camera to obtain a clear image. If multiple images must be collected, this slows the process even further.
Another off-head system uses line scan cameras instead of 2D array cameras, which typically provide better spatial resolution. Compared to area array cameras, line scan cameras will typically have shorter exposure times, hence higher brightness light sources are generally required. In a line scan camera, the image is built-up one line of pixels at a time as the object and the camera move relative to one another.
In order to optically assess the quality of components before placement on a circuit board, several systems allow for inspection of lead coplanarity, missing or improperly placed balls or other quality issues with the component.
One off-head system approach uses standard CCD cameras to image at least a portion of the component, either as the component passes unidirectionally over a line scan camera, or with the component fixedly held ever the camera. From an image of the portion of interest on the component, computations about the lead tweeze in a component with leads, missing balls on a ball grid array or the like may be made. This system cannot assess any parameter related to height such as lead or ball coplanarity.
One off-head sensor system for assessing lead coplanarity is shown schematically in FIG. 4 and described in U.S. Pat. Nos. 5.33 1,406 and 5,309,223 to Konicek and Fishbaine, which pre-aligns the component perpendicularly with respect to the direction of scan. The system employs two or more beams of light which intersect in a point, where the height of the component is held constant with respect to the line while the leads on one side of the component are sequentially passed through the beams. As the component moves at the constant reference height, each lead sequentially blocks and then unblocks at least two of the beams. Once the leads on the first edge of the component are analyzed, leads on the second, third and fourth edges of the component are analyzed.
Another version of this method uses two similar sensors positioned on opposite sides of a rectangular component, which increases throughput by a factor of two, because all four edges of the component are analyzed in two movements of the quill. Because this is a shadow cast system, it is unable to inspect the features on the bottom of components, such as improperly manufactured balls on a BGA.
Another approach uses a high aspect ratio detector (possibly both in terms of pixel count and aspect ratio of individual detectors) that is placed at approximately the same height as the leads to be measured. Two or more point sources are placed opposite the detector and slightly below the leads. Images are collected of the shadows cast from each point source and analyzed to determine the heights of the leads. One version of this has four copies of this geometry, one for each side of the component. This approach is very fast, and has the added advantage of being on-head. With appropriate software it may also be possible to determine the component position and orientation with sufficient accuracy for placement. The disadvantages of the device are its need for a custom detector array and the large size for an on-head device (hence limiting the number of heads which can be installed on a machine).
Another approach is to scan a laser across the device and observe the laser""s apparent position from an oblique angle (which may also be scanned). The primary advantages of this approach are speed and versatility. As this method is a front-lit system, features on the bottom of parts can be measured. A resonant or polygon scanner shifts the beam rapidly across the component in one axis while the component is translated in the other. The disadvantages of this approach are the relative complexity of the sensor, the challenge of dealing with a wide range of surface reflectivities, and maintaining a constant height for the component while scanning.
The prior art has its limitations, and a single system which takes advantage of the existing precision rotary motion capability of the pick and place machine, provides orientation and component quality information, and is capable of providing on-head coplanarity information without requiring pre-alignment steps is needed.
The system of the present invention reports a signal related to a physical condition of an object, such as an electronic component, with the most basic realization of the system including a vacuum quill for releasably holding the object and a motion control system for rotating the quill. The invention includes control electronics coupled to the detector for providing a trigger signal where the detector is oriented to view a stripe in a viewing plane perpendicular to the central axis of the quill, and to provide an image of the stripe. The control electronics sends a plurality of trigger signals to the detector while the motion control system rotates the quill, with each trigger signal triggering the acquisition of another image of a stripe. A processing circuit processes the plurality of images of the stripes to provide the signal related to the physical condition of the object, which can include the orientation or location of the component, the presence or absence of balls on a ball grid array, the height of a specific lead on a leaded component, the distance between the leads on a leaded component or the coplanarity of features on the component.
As required for the particular application of the invention, various sorts of illumination can be provided to create additional embodiments of the present invention, such as front-lit, including specular, diffuse and combinations thereof, back-lit, back-lit shadow casting, reflective shadow casting or a combination of illumination types. The present invention can be integrated into various types of pick and place machines to provide on-head or off-head reporting of the physical condition of interest. The detector is triggered to acquire an image as a function of elapsed time or upon reaching preselected angular positions of the quill. The processing circuit may be configured to operate in real time or on stored images of the stripes. If operating in real-time, the processor locates at least two data points from an array of polar data, and computes the signal as a function of the two or more polar data points. Otherwise, the processing circuit may convert the polar data, collected in a polar array into a Cartesian array, by mapping gray values in the polar array into target pixels of a Cartesian array.
When the detector in the present invention is oriented to view a plurality of stripes radially oriented with respect to the center of the component, the component is rotated over 360 degrees to acquire an image of the entire component. When the detector is oriented to view a cross component stripe, the component need only be rotated over 180 degrees in order to acquire the necessary data to provide the signal. Off-axis configurations of the detector with respect to a central axis of the quill may be employed when the application requires information about the height of features on the component, such as coplanarity information. In order to generate height information, an off-axis geometry is used, a detector oriented to view a cross-component stripe is used, and the component is rotated 360 degrees to obtain two views of the features necessary to compute heights.
Various embodiments of the quill in the pick and place machine, or variations in offset for the quill pickup may be employed with any embodiment of the present invention in order to minimize space or limit the collection of unnecessary data. The pick-up nozzle may have a turn in it so that the center of the component and an axis of rotation of the quill are not coincident. A second axis of rotation near the center of the component may be added that can rotate independently of the primary quill rotation. In a particular embodiment of this, the two axes rotate at substantially equal and opposite rates. In another embodiment, the central axis of the quill is outside of the outline of the component, which provides for faster inspection of larger components. In another embodiment, multiple quills, or multiple nozzles on one quill may be employed. Optional mirrors may be employed in several of the embodiments. One mirror embodiment movably or fixedly places a mirror in the optical path between the source and the component, while another embodiment movably or fixedly places a mirror in the optical path between the component and the detector.
The present invention also includes a method for picking and placing a component, comprising the steps of releasably picking up, the component with a quill, the quill having a central axis perpendicular to a viewing plane, positioning the component relative to a detector, the detector adapted to view a stripe in the viewing plane and including electronics adapted to provide a signal representative of a physical condition of the component and rotating the component while acquiring a plurality of views of the stripes. At this point, the pick and place machine may optionally decide whether to place the component as a function of the signal. In any event, the method continues to move the component to a placement area and finally, place the component as a function of the signal. The method may be practiced on-head or off-head. The present invention also includes a method for picking and placing a component by rotating the component out of the way of the sensor housing when the quill elevates the component, rotating the component to acquire a plurality of stripes according to the apparatus of the present invention, and finally rotating the component out of the way of the sensor housing before lowering the quill on the way to placing the component.